Optimal Head Rotation for Internal Jugular Vein Cannulation When Relying on External Landmarks

Lieberman, Jeremy A. MD; Williams, Kayode A. MD; Rosenberg, Andrew L. MD

Anesthesia & Analgesia:
doi: 10.1213/01.ANE.0000132908.77111.CA
Cardiovascular Anesthesia: Research Report
Chinese Language Editions

External anatomic landmarks have traditionally been used to approximate the location of the neck blood vessels to optimize central venous cannulation of the internal jugular vein (IJV) while avoiding the common carotid artery (CCA). Head rotation affects vessel orientation, but most landmark techniques do not specify its optimal degree. We simulated catheter insertion via both an anterior and central approach to the right IJV using an ultrasound probe held in the manner of a syringe and needle in 49 volunteers. Increased head rotation from 0°, 15°, 30°, 45°, and 60° to the left of midline was associated with higher probability of a simulated needle contacting the IJV and the CCA. For both approaches, the risk of CCA contact was <10% for head rotations of ≤45°. Increased body surface area (BSA) and body mass index (BMI) were associated with more CCA contact at head rotations of 45° or 60°. To optimize IJV contact while reducing the likelihood of inadvertent contact with the CCA, the head should be rotated no more than 30° in patients with high BMI or BSA, but it may be turned to 60° if BMI or BSA is low.

In Brief

IMPLICATIONS: In a study of 49 volunteers, an ultrasound probe held in the manner of a syringe and needle demonstrated that the optimal head rotation to maximize internal jugular vein cannulation and minimize carotid artery puncture was 30&amp;#x00B0; from midline for obese patients and as much as 60&amp;#x00B0; for non-obese patients.

Author Information

Department of Anesthesiology and Critical Care Medicine, University of Michigan Medical Center, Ann Arbor, Michigan

All funding for this study was provided by the Department of Anesthesiology, University of Michigan Medical Center, Ann Arbor, MI.

Accepted for publication May 5, 2004.

Address correspondence to Andrew L. Rosenberg, MD, Department of Anesthesiology, University of Michigan Medical Center, Room 1G323 UH, Box 0048, Ann Arbor, MI. Address e-mail to jliberma@umich.edu.

Article Outline

Central venous cannulation using the internal jugular vein (IJV) is useful for the management of a variety of clinical conditions. Inadvertent puncture of the common carotid artery (CCA) is the most frequent complication of central venous catheter insertion, and may result in central nervous system injury or airway compromise (1,2). The reported incidence of CCA puncture is as frequent as 14% (3–6). In these large clinical series, significant predictors of CCA puncture were operator experience and the CCA positioned directly beneath the IJV.

In the absence of direct visualization of the vessels, external anatomic landmarks are used to predict the location of the IJV and CCA. Various landmark-guided techniques have been described; none has proven superior with regard to the rate of successful IJV cannulation or the incidence of CCA puncture (5,7,8). Most landmark-guided methods dictate that the head should be turned away from the side of the neck being entered, but the degree of head rotation is rarely specified (7). Head rotation does affect the position of the IJV relative to the CCA. The IJV usually lies anterior and lateral to the CCA (9). As the head is rotated away from midline, the IJV becomes more directly anterior to the CCA (10). Extreme head rotation, to 80° or 90°, frequently causes the CCA to sit directly underneath the IJV, increasing the theoretical risk of CCA puncture (11,12).

We attempted to determine whether successful IJV cannulation and the risk of CCA puncture could be altered based solely on varying head rotation. There are no studies specifically evaluating how the degree of head rotation affects the likelihood of a needle entering the IJV or the CCA when relying only upon external anatomic landmarks. Studies in patients involving actual multiple needle insertions into the neck are impractical and unethical. Therefore, we describe a model that simulates needle insertion, using an ultrasound probe oriented in the manner of a syringe and needle. We hypothesized that our simulated needle would more likely be directed toward the CCA for head rotation >30°, thus creating increased risk for CCA puncture as compared with head rotation <30° from the midline.

Back to Top | Article Outline


Patient Selection and Study Design

After approval from our IRB, we obtained the data for this prospective study from a cohort of 49 healthy adult volunteers recruited by poster announcements in a large tertiary medical center between October 2000 and January 2001. All participants gave written informed consent. We excluded individuals with any known abnormal neck anatomy, previous surgery or trauma involving the neck, or prior cannulation of neck vessels.

The primary outcome (dependent) variables were “hits” by a simulated needle to the IJV and CCA. We simulated a needle, extending from the tip of the ultrasound probe and passing through the neck, by using the ultrasound depth gauge to indicate the central 1.5 mm of the ultrasound image along its directional axis (Fig. 1). A hit was defined as the intersection of any part of this central 1.5-mm band and any part of the inner lumen of the IJV and/or CCA (Fig. 1). The predictor (independent) variables were angle of head rotation from midline, central versus anterior approach to the neck vessels, age, gender, body surface area (BSA), and body mass index (BMI). BSA was reported as meters (2) and was calculated using the Mosteller formula [{height (cm) * weight (kg)/3600}1/2] (13). BMI was calculated as the body weight divided by the square of the height (kg/m2). Large BMI was defined as a value >25 kg/m2, which corresponds to the Food and Drug Administration definition of obesity (14). A large BSA was defined for any value above the median (1.87) for our study population.

Back to Top | Article Outline

Each subject was placed supine on a gurney which was then tilted 15° in the Trendelenburg position. Two marks were placed on the skin of the right neck, representing the point where a needle would be inserted to access the IJV (Fig. 2). The locations of the marks are based on two commonly used surface landmark techniques. Simulating one type of the “central” approach (15,16), we placed a mark at the apex of the triangle formed by the two heads (sternal and clavicular) of the sternocleidomastoid muscle (SCM) (Fig. 2A). We placed a mark on the skin at the level of the cricoid cartilage, along the medial edge of the SCM, to simulate the “anterior” approach (4). No interventions to increase the diameter of the neck vessels, such as a Valsalva or breath-holding maneuver, were performed by the subjects.

The locations of the CCA and IJV were observed using a portable two-dimensional ultrasound imager (Site-Rite®II Ultrasound Scanner; Dymax Corporation, BARD Access Systems, Pittsburgh, PA). The probe was held in the manner of a syringe and needle to simulate a typical needle-insertion angle used in central line insertion. The ultrasound probe was held within a parasagittal plane and directed 30° caudad (Fig. 2, B and C). The probe was held to the skin with minimal pressure to limit neck vessel compression (3). A single, experienced investigator (KW) placed all surface markings and manipulated the ultrasound probe. After the probe was placed in proper position, another investigator (JL) collected and analyzed the ultrasound data in real time whereas the investigator holding the probe was kept blinded to the findings. All measurements were performed using the skin marks for both the anterior and central landmark approaches. These observations and measurements were repeated as the subject’s head was rotated through 5 positions: 0° (head midline), 15°, 30°, 45°, and 60° to the left of midline (Fig. 2D). The external anatomic landmarks were reevaluated at each head position and adjustments to the insertion marks were made, if necessary. A carpenter’s level was used to ensure that the short axis of the gurney surface was parallel to the floor without any lateral tilt or list and that the long axis of the gurney was tilted 15° in the Trendelenburg position. Head rotation angle of 0° was defined as having the subject’s sagittal plane perpendicular to the floor. We used a leveled protractor to align the sagittal plane to the desired angle (Fig. 2D). Similarly, we attached a level to the ultrasound device to avoid any medial or lateral deviation of the beam. A leveled protractor was also attached to consistently direct the beam 30° caudad.

For analyses involving the IJV, the proportion of hits was measured for each method and for each angle of head rotation. A logistic regression model was used to analyze differences between head angles, adjusting for repeated measures on the same subject using Generalized Estimating Equations methodology. The effects of age, gender, large BSA, and large BMI were estimated in the logistic regression. These variables were chosen a priori because they had been found in previous studies to be associated with alterations in neck anatomy and jugular and carotid location after moving the head/neck (9,11,12). Because of the small number of hits for the CCA, it was not possible to conduct a logistic regression to compare angles. Rather, a contingency table analysis was performed, again, adjusting for multiple measurements made on the same subject. Fisher’s exact test was used to compare the proportion of hits by gender, large BSA, and large BMI. On the basis of results from previous studies (10,11), we determined that 25 patients would be required for each approach (anterior and central) in order to detect an absolute increase in CCA hit rate of at least 25% from an estimated 14% to 39% when the head was moved beyond 30° (comparison of matched proportions, assumed discordance = 25%, α = 0.05, β = 0.2). All analyses were performed using the computer software SAS release 8.2 (copyright 1999–2001; SAS Institute Inc., Cary, NC) except the contingency tables which were performed using the software STATA version 6.0 (copyright 1984–1999; STATA Corporation, College Station, TX), and the sample size estimate using nQuery Advisor 4.0; Statistical Solutions, Ltd., Cork, Ireland).

Back to Top | Article Outline


Data were collected for all 49 patients enrolled in the study. There were no complications or adverse events in any of the subjects. Demographic data are shown in Table 1. There was an equal number of men and women in the study. Men were statistically taller, heavier, and had larger BSAs than women. The BMIs were equal for men and women (Table 1).

Both the IJV and CCA were present and patent by ultrasound visualization in all subjects for all head positions. Figure 3 depicts the frequency with which the simulated needle hit the IJV or CCA (as defined in Methods) as the subjects’ heads were rotated through 5 positions between 0° (midline) and 60°. The degree of head rotation was significantly associated with the ultrasound beam intersecting the IJV using either the anterior or central approach (P < 0.001). However, there were no significant differences in IJV intersection between the two approaches. The probability of hitting the IJV using the anterior approach increased for each increment of head rotation from 0° through 60° (P < 0.03). The central approach improved IJV contact for head rotation of ≥30° versus 0° or 15° (P < 0.001–0.005, respectively), but there was no significant difference in IJV hits between 30° and 60°.

The simulated needle did not hit the CCA until the head was rotated at least 30°. As the head position was rotated further, the CCA hit rate was higher at 45° and 60° compared to 30° for both approaches (P < 0.01).

Age and gender did not affect the likelihood of hitting the IJV or CCA at any amount of head rotation. Large BMI or large BSA did not alter the association of head rotation to hitting the IJV. Large BSA and large BMI were independently associated with an increased risk of a simulated needle hitting the CCA when the head was rotated to 45° or 60°. The incidence of CCA hits for patients with large BSA and large BMI was more frequent using the central approach versus the anterior approach (P < 0.05). For subjects with small BSA or small BMI, the incidence of CCA contact was ≤4% for all head positions, using either approach.

Back to Top | Article Outline


Our study is the first human trial to specifically address the consequence of head rotation upon the success of attempted venous cannulation; albeit, using an ultrasound probe to simulate an actual syringe and needle. Our data complement and extend the findings of previous studies (3,6,7,9) by demonstrating that, when directed according to established surface landmark techniques, progressive rotation of the head away from the midline position increases the probability that a needle will intersect the IJV. However, we found no statistical difference in IJV contact when using an anterior versus a central approach. This supports the findings of Metz et al. (7), who used similar methodology and found no advantage among numerous anterior or central approaches. However, they evaluated each technique with the head rotated in a single position. We found that both approaches equally predict IJV contact over a wide range of head positions. Differences in age, gender, BSA, and BMI did not influence the effect of head rotation on the probability of hitting the IJV in our population.

In all subjects, and for both approaches, our data indicate that the likelihood of a needle hitting the CCA is minimal if the head is turned ≤30° from midline. Previous studies have suggested that head rotation away from the midline increases the risk for CCA puncture (9,11,12). However, these observations were based only on ultrasound images demonstrating increased vessel overlap. Metz et al., using similar methodology, reported that the incidence of IJV and CCA hits was not affected by head rotation. Our analysis of their primary data (Table 2 of their results) shows a more frequent CCA hit rate in both central and anterior approaches that advocated maximal (90°) head rotation versus minimal (30°) or zero rotation (34% versus 16% versus 1.5%, respectively) (7). Their finding of no statistically significant effect of head rotation may be attributed to both small sample size and related to other variations mandated by the particular techniques tested.

In this study, the risk of contact with the CCA increased significantly only in subjects with larger BSA and BMI as the head was turned to 45° or 60°. This effect is not likely to be attributed to larger patients having larger caliber carotid arteries because other investigators have found no relationship between arterial size and height or weight (9). Compression of the IJV cannot explain our findings, as it does not occur unless there is extreme head rotation (approximately 90°), which we avoided (11). We speculate that larger and heavier people perhaps have a larger SCM. As the head is rotated, the skin entry points shift medially, regardless of whether the landmark technique relies on the medial border of the SCM (anterior) or the apex of the triangle formed by its two heads (central). Our experience indicated that these changes in entry point with head rotation were minimal. However, it remains possible that larger muscle size may shift the skin entry site more medially during head rotation, directing the needle further toward the CCA. Further studies may also help elucidate whether there are clinically significant differences among non-obese patients with muscularly thick necks to see if these relationships hold in these different populations.

Extending our data to an actual clinical situation should be done with some caution. Troianos et al. (9) found that the degree of overlap of the CCA and IJV is significantly greater in subjects older than 60 years. In our study population, all subjects were <60 years old; hence, generalizing our study’s data to older patients may be limited. More importantly, we used an ultrasound probe applied to the skin to simulate a syringe and needle cannulating the IJV. Although we blinded the probe operator to the results of intersecting a vessel or not, there is a possibility of observer bias being introduced by not having a blinded reviewer analyze photos of the ultrasound images. We did not do this because we wanted to exclude the effects of subtle changes in vessel variables during breathing, patient movement, or operator movement. Photographs of our ultrasound images would have avoided potential bias, although the exact timing of when a picture would be taken would still be subject to operator bias. In addition, we believe that our definitions of hit or miss were of sufficient objectivity to minimize the effect of observer bias.

Also, a clinical trial involving actual needle sticks at five different head rotations would be impractical. Moreover, it is unethical to allow an operator to attempt cannulation if ultrasound assessment suggested the approach presented a high risk for CCA puncture. Ultrasonography has been previously used to demonstrate the anatomic relationship between the IJV and CCA (9,11,12,17). These studies characterized the risk for CCA puncture based on the relative position of the IJV to the CCA. They found that the risk of CCA puncture was more likely if the IJV was not visualized, if the IJV was not anterolateral to the CCA, or if there was overlap of the two vessels. However, overlap of the vessels may not accurately predict true risk for CCA puncture because the vein will almost always be entered before the artery. However, the vein may be collapsed by the needle, increasing the risk of a needle entering the artery if it is directly below the vein. Therefore, our model is somewhat limited in that we cannot predict how an actual needle will be affected by contact with tissue such as displacing the vessel rather than entering the lumen. Factors such as the bevel of the needle, the pressure within the vessels, presence of thrombus and arterial wall calcification may influence whether an actual needle enters a vessel, but are beyond the scope of this study. Although our method attempts to be more realistic by assessing whether a simulated needle actually intersects one or both vessels, it may be overestimating the risk of carotid puncture.

We evaluated one anterior and one central approach based on previously published descriptions. These techniques are based on easily identifiable external landmarks that have been previously used in large patient series. This was done to compare our surrogate predicted outcomes with clinical outcomes previously demonstrated (3,7,8). For both anterior and central approaches, the ultrasound beam was directed caudad, along a parasagittal plane. Using the sagittal plane as a reference point reduces the gap between probe’s lateral face and the skin, providing an optimal interface at potentially lower skin pressures. This was likely to reduce the distortion of the neck vessels from the probe and improve simulating the results when using a syringe and needle. We also chose approaches with no medial or lateral deviation to improve reproducibility, as multiple measurements were performed on each subject. Clearly, this does not allow us to extend our findings to techniques that recommend that needle insertion be directed laterally or toward the ipsilateral nipple. Another limitation of our study may be that identifying the apex of the SCM triangle and cricoid cartilage may have been influenced by operator experience. In obese people, these landmarks may be difficult to find. We attempted to avoid the confounding effect of operator experience by relying on a single, experienced operator to determine all landmarks. Finally, there is a limitation of using any simulation technique as compared with an actual clinical study. Validating the methodology used in our study might be both difficult, but also important for further studies that use ultrasound probe simulation. After determining, by ultrasound, whether a needle would hit the vessel, the first finder needle pass of the actual attempt at right IJV cannulation could be performed by a mechanical device applying a standardized needlestick in a predetermined direction to a predetermined depth at only one of the safe angles of head rotation per patient. A stopcock could be attached to this system that would allow sampling the blood Pao2 to determine if there is only venous blood or if arterial blood is also present as a means to detect arterial puncture.

We only examined head rotation ranging from 0° to 60°. We did not exceed 60° because previous studies reported significant vessel overlap at 80° and 90° (9,11,12). We chose 15° increments of head rotation because an operator can reasonably approximate angles of 30°, 45°, or 60° without needing tools. A simple clinical guideline may be the following: Angles of 0°–90° are easily identified. Therefore, 45° is half the deviation between the 2 extremes whereas 30° and 60° are a bit less or more than 45°, respectively. We limited our study to only the right side of the neck because cannulation of the right IJV is often preferred to cannulating the left (8,15). This is because the path to the right atrium is more direct, the dome of the right pleura is lower, and the thoracic duct is on the left. Moreover, in a randomized series, the incidence of complications is more frequent when cannulating the left IJV versus the right when relying solely on landmarks (18). The anatomic relationship of the IJV to the CCA does differ between right and left (10,12). Our data may not be as applicable when attempting to cannulate the left IJV.

If maximizing the likelihood of entering the IJV were our sole objective, then we would have concluded that the best head rotation was 45° or 60°. This amount of head rotation yielded the most IJV hits ranging from 52% to 68% in our population. This is similar to results from actual clinical series (3,4). However, our data also showed that the risk of a needle entering the CCA also increased as the head was turned laterally. To define an optimal angle of head rotation, we must decide what risk of CCA contact is acceptable. We conclude that a probability of the ultrasound beam intersecting the CCA of no more than 10% is permissible, as this actually overpredicts the true risk of CCA puncture. In most of our subjects, when the CCA was hit, the ultrasound beam passed through the IJV first. Thus, many of the potential CCA punctures would be avoided if needle advancement ended at the IJV, before reaching the CCA. However, passing through the IJV first does not preclude CCA puncture, as the back wall of IJV may be penetrated in up to 50% of IJV cannulations (19). This would imply a risk of CCA puncture of about 5% which is consistent with actual CCA puncture rates of 4.2%–8.3% obtained in large clinical trials relying on landmark-guided techniques (3,4,6,20).

Applying the above definition for optimal head rotation, we suggest that, in general, the head should not be turned beyond 45°. In subjects with large BSA or large BMI, optimal rotation may be achieved at 30°, whereas subjects with small BSA or small BMI have an optimal head rotation of 45° or 60°.

In summary, although ultrasound-guided placement of central venous catheters improves success rates and decreases carotid puncture compared with use of external landmark techniques alone (6,21–23), surface landmark techniques remain a common localization method. This may be attributed to limitations using the probes such as the cost of portable ultrasound devices, lack of training in how to use these devices, and their relative lack of availability in multiple venues. Central venous catheters are often placed in an operating room or intensive care unit where the ultrasound probes are not available or not working properly because the batteries have lost charge. Finally, using a probe can be somewhat cumbersome while maintaining sterility. When one must rely on surface landmark guidance, our study facilitates choosing a head position that may help to maximize IJV entry yet reduce CCA puncture. Validation of our simulation model and similar assessment of other cannulation techniques are indicated. Additional studies in older patients and those with cardiopulmonary disease may be helpful to evaluate if these comorbidities change what appears to be the optimum angle of head rotation for IJV cannulation.

The authors are indebted to Kathleen Welch, MPH, MS (Center for Statistical Consultation and Research, Ann Arbor, MI), for her invaluable help with statistical analysis.

Back to Top | Article Outline


1. Heath KJ, Woulfe J, Lownie S, et al. A devastating complication of inadvertent carotid artery puncture. Anesthesiology 1998;89:1273–5.
2. Anagnou J. Cerebrovascular accident during percutaneous cannulation of internal jugular vein. Lancet 1982;ii:377–8.
3. Denys BG, Uretsky BF, Reddy PS. Ultrasound-assisted cannulation of the internal jugular vein. Circulation 1993;87:1557–62.
4. Goldfarb C, Lebrec D. Percutaneous cannulation of the internal jugular vein in patients with coagulopathies: an experience based on 1000 attempts. Anesthesiology 1982;56:321–3.
5. Sznajder JI, Zveibil FR, Bitterman H, et al. Central vein catheterization: failure and complication rates by three percutaneous approaches. Arch Intern Med 1986;146:259–61.
6. Troianos CA, Jobes DR, Ellison N. Ultrasound-guided cannulation of the internal jugular vein: a prospective, randomized study. Anesth Analg 1991;72:823–6.
7. Metz S, Horrow JC, Balcar I. A controlled comparison of techniques for locating the internal jugular vein using ultrasonography. Anesth Analg 1984;63:673–9.
8. Defalque RJ. Percutaneous catheterization of the internal jugular vein. Anesth Analg 1974;53:116–21.
9. Troianos CA, Kuwik RJ, Pasqual JR, et al. Internal jugular vein and carotid artery anatomic relation as determined by ultrasonography. Anesthesiology 1996;85:43–8.
10. Dickson CS, Roth SM, Russell JM, et al. Placement of internal jugular vein central venous catheters: anatomic ultrasound assessment and literature review. Surgical Rounds 1996;3:102–7.
11. Bazaral M, Harlan S. Ultrasonographic anatomy of the internal jugular vein relevant to percutaneous cannulation. Crit Care Med 1981;9:307–10.
12. Sulek CA, Gravenstein N, Blackshear RH, Weiss L. Head rotation during internal jugular vein cannulation and the risk of carotid artery puncture. Anesth Analg 1996;82:125–8.
13. Mosteller RD. Simplified calculation of body-surface area. N Engl J Med 1987;317:1098.
14. Anonymous. Executive summary of the clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. Arch Int Med 1998;158:1855–67.
15. Daily PO, Griepp RB, Shumway NE. Percutaneous internal jugular vein cannulation. Arch Surg 1970;101:534–6.
16. Civetta JM, Gabel JC, Gemer M. Internal-jugular-vein puncture with a margin of safety. Anesthesiology 1972;36:622–3.
17. Denys BG, Uretsky BF. Anatomic variations of internal jugular vein location: impact on central venous access. Crit Care Med 1991;19:1516–9.
18. Sulek CA, Blas ML, Lobato EB. A randomized study of left versus right internal jugular vein cannulation in adults. J Clin Anesth 2000;12:142–5.
19. Mangar D, Turnage WS, Mohamed SA. Is the internal jugular vein cannulated during insertion or withdrawal of the needle during central venous cannulation? [letter]. Anesth Analg 1993;76:1375.
20. Jobes DR, Schwartz AJ, Greenhow DE, et al. Safer jugular vein cannulation: recognition of arterial puncture and preferential use of the external jugular route. Anesthesiology 1983;59:353–5.
21. Legler D, Nugent M. Doppler localization of the internal jugular vein facilitates central venous cannulation. Anesthesiology 1984;60:481–2.
22. Mallory DL, McGee WT, Shawker TH, et al. Ultrasound guidance improves the success rate of internal jugular cannulation: a prospective, randomized trial. Chest 1990;98:157–60.
23. Randolph AG, Cook DJ, Gonzales CA, et al. Ultrasound guidance for placement of central venous catheters: a meta-analysis of the literature. Crit Care Med 1996;24:2053–8.
© 2004 International Anesthesia Research Society